AI and brainstorming are often seen as major innovation levers. Yet our data analysis (from hundreds of brainstorming sessions since 2002) and tests conducted with 12 AIs show that these approaches don't always deliver on their promises.
Brainstorming Limits for Innovation
Since 2002, we have been analyzing creative processes in R&D. While brainstorming is useful for generating ideas quickly, it has clear limitations:
- Very few truly innovative ideas (brainstorming was never designed for that)
- Dynamics influenced by human biases (dominant personalities, social pressure)
- Sometimes lengthy discussions, for often predictable results
- Ideas dismissed prematurely, due to lack of consensus or a suitable framework
23 Years of Brainstorming vs 12 AIs
To objectively compare brainstorming and AI, we submitted the same technical problem to 12 artificial intelligences, then compared their responses to the results observed during hundreds of brainstorming sessions since 2002.
AI or Brainstorming: which will perform better?
AIs have encyclopedic knowledge, impressive synthesis capabilities and unmatched speed. Brainstorming, on the other hand, mobilizes field experience and collective intelligence. Let's see what each one proposes when facing a concrete problem.
Study Subject: the Antenna Problem
Mission: espionage and autonomous transmission. Carried by one person. Extreme conditions (mountains, cold).
Critical Issue
Ice accumulates on the antenna rake. Beyond a certain threshold, the mast breaks under the weight.
Question asked:
How can this problem be fixed for the next generation of antennas?
Solutions Proposed in Brainstorming
Hundreds of workshops, the same patterns observed systematically.
List of ideas in order of appearance:
- Reinforce the mast
- Heat or vibrations
- Cables/guy-wires, tripods
- Flexible/articulated/telescopic mast
- Surface coating (Teflon)
- Change of shape or orientation
- Tent or roof
- Balloon (replace the mast)
- Change technology (satellite, drone)
- Go clean the antenna
- Accelerate global warming
- Stop the war
- Train animals
A Predictable Progression
Ideas move from simple to complex, gradually drifting away from the initial problem. This is a classic cognitive mechanism: the brain first favors immediate solutions, then drifts toward increasingly remote approaches.
During a workshop in Toulouse with 500 participants, the list of proposals reproduced this exact progression.
Solutions Proposed by 5 AIs
We added the instruction "make a numbered list of 10 effective ideas." Here are the 5 most representative AIs.
1. ChatGPT - OpenAI
(GPT-5, initially refused the topic for military reasons)
- Heating system integrated into the rake (low-consumption electric resistors)
- Hydrophobic anti-adhesion coating (Teflon or nano-coating type)
- Mast made of shape-memory composite material
- Periodic vibration system to dislodge ice
- Retractable rake (folds when weight exceeds a threshold)
- Inflatable structure replacing the rigid mast
- Load sensor with alert and automatic degraded mode
- Telescopic mast with adjustable height
- Aerodynamic design limiting wind and ice exposure
- Dual mast with automatic switchover to the second
2. Grok - xAI
- Autonomous electric resistance heating
- Super-hydrophobic coating (nano-structured)
- Ultrasonic vibrations for de-icing
- Reinforced carbon fiber mast
- Automatic folding mechanism under load
- Wire antenna stretched between natural supports
- Piezoelectric coating generating heat
- Profiled design reducing accumulation
- Relay drone replacing the fixed antenna
- Integrated phase-change material
3. Le Chat - Mistral AI
- Low-power electric heating on the rake
- Hydrophobic or icephobic coating
- Variable cross-section mast (thicker at the base)
- Mechanical vibration system
- Adjustable rake facing into the wind
- Deployable reinforcement guy-wires
- High buckling-resistance composite material
- Articulated structure absorbing constraints
- Frost sensor with automatic retraction
- Aerodynamic rake shape limiting ice buildup
4. Gemini - Google
- Battery-powered heating resistors
- Anti-frost coating (nanotechnology)
- Reinforced mast in high-performance alloy
- Controlled periodic vibrations
- Retractable telescopic mast
- Rotating rake preventing accumulation
- Deployable physical protection (fairing)
- Flexible wire-based antenna
- Overload detector with operator alert
- Modular design with quick-swap spare parts
5. Claude - Anthropic
(Sonnet 4)
- Low-consumption integrated heating
- Advanced hydrophobic/icephobic coating
- Composite mast with controlled flex zone
- Vibration-based de-icing system
- Retractable rake under load threshold
- Mast-free antenna architecture (wire between high points)
- Phase-change material on the surface
- Aerodynamic profile reducing ice buildup
- Strain sensor with preventive retraction
- Mast with optimized conical cross-section
Findings on AI Results: a Glaring Lack of Innovation
None of the solutions proposed by AIs break out of the classic brainstorming framework. All follow a linear causal approach, with 4 recurring strategies: prevent ice buildup, reduce adhesion, remove after accumulation, reinforce the structure.
The Causal Approach Stifles Innovation
Both AI and brainstorming reason the same way: identify the cause (ice), then act on it. This linear reasoning excludes by design any solution that does not target the cause. Yet that is precisely where breakthrough ideas are found.
Looping Responses, No Breakthroughs
All 12 AIs tested recycle the same categories of solutions, with variations in wording. We find exactly the same families of ideas as in brainstorming sessions. The technology changes, the result stays the same.
Heating the Antenna: a Terrible Idea, Yet the Champion
Heating the rake, a bad idea that always comes first
It is the worst solution, and yet it is systematically proposed first, by humans and AIs alike. Heating is heavy, fragile, energy-intensive, and off-target (the antenna must remain autonomous in the mountains).
Why do we adopt it? By reflex (it freezes, so we heat), without questioning the real need.
Yet Other Ideas Exist
1) The mast-free antenna: an organic assembly protocol
Remove the mast. Integrate the antenna into the environment: a tree, a rock, the natural terrain. Advantages: zero exposed structure, natural camouflage, simplified logistics.
2) Foam sleeve or structural spikes: turning ice into armor
Cover the mast with foam or spikes for controlled ice accumulation. Instead of fighting the ice, we exploit it: it reinforces the structure instead of destroying it.
These ideas (and others) are not generated by AIs or brainstorming sessions. How do you evaluate an idea you never had?
Why Do These Ideas Escape Both AI and Brainstorming?
The structural biases are the same: linear causal reasoning, assumptions never questioned, no tool to trigger a paradigm shift. Both AI and brainstorming remain locked within the frame of the problem as stated.
The Solution? A Systematic Approach to Innovation: ASIT
The ASIT method forces lateral thinking through a reproducible process. It does not ask you to be creative: it mechanically generates avenues that neither intuition nor AI can find. Idea evaluation is objective, structured, independent of the author's charisma.
The ASIT Method
ASIT is not a brainstorming session. It is a systematic method that requires rigorous preparation and produces around 30 questions per workshop, each designed to force a perspective shift.
ASIT: The Art of Asking the Right Questions
"How could ice prevent the mast from breaking?"
"How could removing the mast guarantee the system works?"
Why Does It Work?
No creative pressure: you answer precise questions, not "come up with ideas." Hidden opportunities become obvious. The process is reproducible: same problem, same questions, same results, regardless of the group.
Going Further
To discover how to apply this approach to your innovation challenges:
- Our structured creativity training programs
- Creativity and innovation workshops
- Innovation webinars
- The ASIT method in detail
- Article: You "innovate" with AI? LOL!
Innovation is not born from inspiration, but from the right questions.
